OpenAI just launched o3-pro, its most capable AI model yet. It’s a step up from the earlier o3 reasoning model, built to solve problems step by step. That means better performance in tough areas like physics, math, and coding.
O3-pro replaces o1-pro for ChatGPT Pro and Team users starting Tuesday. Enterprise and Edu users get it next week. The model is also live on OpenAI’s developer API as of today.
API pricing: $20 per million input tokens, $80 per million output tokens. For scale, a million input tokens equals about 750,000 words — a bit longer than War and Peace. 
OpenAI tweeted the rollout:
OpenAI o3-pro is available in the model picker for Pro and Team users starting today, replacing OpenAI o1-pro.
Enterprise and Edu users will get access the week after.
As o3-pro uses the same underlying model as o3, full safety details can be found in the o3 system card.…
Reviewers consistently prefer o3-pro over o3 across science, education, programming, business, and writing tasks. It scores higher on clarity, comprehensiveness, instruction-following, and accuracy.
O3-pro comes with tools for web search, file analysis, visual input reasoning, Python coding, and personalized responses with memory. The catch: it takes longer to respond than o1-pro.
Limitations remain: Temporary chats in ChatGPT are disabled due to a “technical issue.” No image generation. Canvas, OpenAI’s workspace feature, isn’t supported yet.
In benchmarks, o3-pro beats top-tier rivals. It edges out Google’s Gemini 2.5 Pro on AIME 2024 (math skills). It also outperforms Anthropic’s Claude 4 Opus on GPQA Diamond, which tests PhD-level science knowledge.
